do memory cd4 t cells keep their cell-type programming ... · cellular reprogramming techniques,...

Do Memory CD4 T Cells Keep Their Cell-TypeProgramming: Plasticity versus FateCommitment?

Epigenome: A Dynamic Vehicle for Transmittingand Recording Cytokine Signaling

John L. Johnson1,2 and Golnaz Vahedi1,2

1Department of Genetics, Perelman School of Medicine, University of Pennsylvania, Philadelphia,Pennsylvania 19104

2Institute for Immunology, Perelman School of Medicine, University of Pennsylvania, Philadelphia,Pennsylvania 19104

Correspondence: [email protected]

CD4þ T cells are critical for the elimination of an immense array of microbial pathogens.Although there are aspects of helper T-cell differentiation that can be modeled as a classiccell-fate commitment, CD4þ T cells also maintain considerable flexibility in their transcrip-tional program. Here, we present an overview of chromatin biology during cellular repro-gramming and, within this context, envision how the scope of cellular reprogramming maybe expanded to further our understanding of the controversy surrounding CD4þ T lympho-cyte plasticity or determinism.

OUTLINE

HOW IS CELL FATE DETERMINED?, 2

HOW FIXED IS THE DIFFERENTIATED STATE?, 2

CELL FATE IN CD4þ T CELLS, 3

REGULATION OF PLASTICITY IN CD4þ T CELLS, 3

INTEGRATING GENOMIC DATASETS FOR A BETTERUNDERSTANDING OF SUBSETS, 4

SINGLE-CELL HETEROGENEITY, 5

LINEAGE TRACING IS IMPORTANT IN DETERMINING PAST, PRESENT,AND FUTURE, 5

CONCLUDING REMARKS, 5

ACKNOWLEDGMENTS, 6

REFERENCES, 6

Copyright # 2018 Cold Spring Harbor Laboratory Press; all rights reserved

Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a028779

1

This is a free sample of content from Immune Memory and Vaccines: Great Debates. Click here for more information on how to buy the book.

© 2018 by Cold Spring Harbor Laboratory Press. All rights reserved.

mailto:[email protected]




http://www.cshperspectives.org/site/misc/terms.xhtml

http://cshlpress.com/default.tpl?action=full&src=pdf&--eqskudatarq=1166

The genetic information encoded in DNA ispresented in the context of chromatin, a

complex of nucleic acid and associated proteinscalled histones. The dynamic nature of the chro-matin is a defining feature of the epigenetic reg-ulation. The regulation of CD4þ T cells similarto any other cell type is therefore tightly linkedwith the epigenome—the combinatorial vari-ance in localization and modifications of chro-matin factors and underlying DNA. Propelledby rapid technological advances in sequencing,the field of epigenomics is enjoying unprece-dented growth with no sign of deceleration (Ri-vera and Ren 2013). An expanding number ofimmunologists are working to explore excitingfrontiers in epigenomic regulation of immunecells. Consequently, the number of data sets andpublications has grown in recent years. Thetheme emerging from these genome-scale stud-ies is that the chromatin structure is an attractivetemplate for both transmitting and recording im-munological events. For example, the demeth-ylation of the PD-1 promoter is recorded duringthe effector phase of T-cell exhaustion (Young-blood et al. 2011), whereas cytokine signaling inCD4þ T helper cells can be transmitted throughgain and loss of histone acetylation at thousandsof genomic regions (Vahedi et al. 2012, 2015).Here, we aim to discuss the controversy sur-rounding CD4þ T lymphocyte plasticity or de-terminism mostly from a chromatin biologyperspective. This article has a three-prong focus:We start by discussing how cell fate is regulatedduring development. We then focus on repro-gramming studies and the characteristics of fac-tors capable of inducing cell-fate changes. Andfinally we evaluate CD4þ T-cell plasticity fromthe chromatin biology perspective.

HOW IS CELL FATE DETERMINED?

Cellular-state information between generationsof developing cells is propagated through theepigenome particularly via the accessible partsof the chromatin. Consistent patterns of gainand loss of chromatin accessibility has been re-ported as cells progress from embryonic stemcells to terminal fates (Stergachis et al. 2013).Chromatin accessibility together with covalent

modifications of histones is controlled in largepart by the action of multiple transcription fac-tors that recognize and bind specific sequencesin the genome. Rather than one regulatory pro-tein being singularly responsible for creating thechromatin state and determining each cell type,particular combinations of transcription fac-tors elicit cell-fate changes and maintain cellidentity (Iwafuchi-Doi and Zaret 2016). Theemerging theme from recent studies is that thechromatin state has a hierarchical organizationwhere every layer is regulated by distinct groupsof transcription factors. The first layer relates toa small number of transcription factors who actfirst on the chromatin and are referred to as“pioneer transcription factors.” These factorsare distinguished from other transcription fac-tors by their ability to bind their cognate DNAsites directly on the nucleosome, even in chro-matin that is locally compacted by linker his-tone (Zaret and Mango 2016).

Such pioneer transcription factors initiate co-operative interactions with other transcriptionfactors to elicit changes in local chromatin struc-ture. A second layer in this hierarchical organiza-tion consists of transcription factors such as AP-1family members JUNB, BATF, and IRF4, whichfurther prime parts that later become associatedwith more specific and dynamic factors (Garberet al. 2012; Kurachi et al. 2014). The bottom layer,particularly essential for CD4þ T-cell biology, re-lates to transcription factors downstream of sig-nal transduction pathways. These signal-depen-dent transcription factors are dynamic andcontrol more specific sets of genes that have com-mon biological functions. For instance, the signaltransducers and activators of transcription (STAT)proteins target the late induced T-helper-spe-cific chromatin states, while the nuclear factorkB (NF-kB) factors Rel, Relb, and NFKB1 targetthe inflammatory program (Garber et al. 2012;Vahedi et al. 2012; Ostuni et al. 2013).

HOW FIXED IS THE DIFFERENTIATED STATE?

The prevailing paradigm in cell developmenthas been that somatic cells become irreversiblycommitted to their fate and lose potency as theyspecialize. Despite physiological settings, exper-

J.L. Johnson and G. Vahedi

2 Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a028779




iments performed several decades ago showedthat the fusion of different pairs of cell typescan awaken dormant gene-expression programs(Blau 1989). Subsequently, breakthroughs incellular reprogramming techniques, recognizedby a Nobel Prize, convinced us that plasticitycan be induced and “fate” can be changed. Re-programming studies on the use of transcrip-tion factors to interconvert cells of differenttypes brought to light the distinguished role ofpioneer factors. Direct assessments of the initialbinding of the induced pluripotency factorsOct4, Sox2, Klf4, and c-Myc showed that Oct4and Sox2, and to a lesser extent Klf4 target pre-dominantly silent and inaccessible chromatin—resistant to DNaseI—when they are forcibly ex-pressed in human fibroblast cells (Soufi et al.2012). Studies in immune cells also showedthat when C/EBPa is ectopically expressed inpre-B cells, it can convert B cells to macrophagesby targeting closed chromatin (van Oevelen etal. 2015). Together, cell fate can be altered ex-perimentally as a result of forced expression ofvarious combinations of transcription factorstypically including pioneer factors.

CELL FATE IN CD4þ T CELLS

We have reviewed concepts in development andreprogramming to highlight that CD4þ T cells,like all other cells, are packaged into chromatin.Specialized CD4þ T cells (e.g., T helper 1 cells),maintain their fate until environmental changeslead to the expression of “influential” transcrip-tion factors that can potentiate cells to changetheir program. Given the stability of the differ-entiated state in vivo, an understanding of theregulation of CD4þ T-cell differentiation bymechanisms that allow the type of plasticity ob-served in reprogramming is critical for our un-derstanding of vaccines and immune memory.

Early in vitro studies argued that once aCD4þ T cell had chosen its fate, it would noteasily switch to another fate, even if exposed tothe cytokines that drove differentiation to theopposing subset (Mosmann and Coffman 1989;O’Shea and Paul 2010; Yamane and Paul 2012).This dogma has been challenged over the lastdecade because the possibility of polarized T cells

to repolarize toward mixed or alternative fateshave been reported (DuPage and Bluestone2016). For example, recent lineage-tracing sys-tems in mice showed that endogenously polar-ized CD4þ T cells from many subsets can altertheir phenotype during their life span (Zhou etal. 2009b; Hirota et al. 2011; Wilhelm et al. 2011;Ahlfors et al. 2014). Furthermore, the pheno-typic plasticity of regulatory T cells that mirrorseach T helper cell subset supports a hypothesisof an inherent flexibility of T cells, both inflam-matory and regulatory, to adapt their functionto changing environments (DuPage and Blue-stone 2016). These in vitro and in vivo studies,extensively reviewed elsewhere (Wilson et al.2009; Zhou et al. 2009a; Murphy and Stockinger2010; O’Shea and Paul 2010; DuPage and Blue-stone 2016), suggest that CD4þ T cells areadaptable and can show phenotypic plasticityin response to changing contexts.

REGULATION OF PLASTICITY IN CD4þ

T CELLS

In this perspective, we propose that plasticity ofCD4þ T cells works by the same mechanismthat initially selected the first fate. If a cell pop-ulation is able to respond to new signals becauseit expresses that specific extracellular receptor,has the cytoplasmic means of signaling, and theability to induce the transcription factors thathave strong impact on the chromatin, then thecellular identity at the epigenetic level can bemodified to allow the cell to take on new iden-tity and function.

The CD4þ T cell’s initial activation throughT-cell receptor (TCR) results in the activation ofprimed elements rather than selection of newregions (Allison et al. 2016). The presence of po-larizing cytokines initiates the process of select-ing the T-cell fate with genome-scale epigeneticchanges (Ciofani et al. 2012; Vahedi et al. 2012).In the face of a new challenge or extracellularenvironment, the presence of new cytokine sig-nals is propagated through available cytokine re-ceptorsandcytoplasmicsignalingevents intothenucleus. These events induce expression of tran-scription factorswithvaryingdegreesof intrinsicability to modify the chromatin environment

Do Memory CD4 T Cells Keep Their Cell-Type Programming?

Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a028779 3




(Vahedi et al. 2012). Although branding tran-scription factors to “master” regulators attracteda lot of attention among T-cell biologists(Rothenberg 2014), we propose to set aside thesemantics and what the words “pioneer” or“master” may imply. Identifying T-helper-spe-cific transcription factors with an intrinsic abil-ity to bind silent chromatin and consequentlyalter T-cell fate has an important implicationfor understanding T-cell plasticity.

INTEGRATING GENOMIC DATASETS FORA BETTER UNDERSTANDING OF SUBSETS

Deciphering mechanisms of CD4þ T-cell plas-ticity requires a more precise definition of these

same subsets (Fig. 1). We argue that this is be-cause populations of cells may appear to fit intoa certain subset based on traditional measures(i.e., flow cytometry), but is actually composedof cells with different histories of cytokine ex-posure that has been imprinted onto the chro-matin. These differences may not be immedi-ately apparent, but may have major influence oncellular function in the right context. Highlyparametric multidimensional measurements ofT cells by mass or flow cytometry has revealedan overwhelming number of subpopulations inany given subset during an immune response(Newell et al. 2012; Gaudilliere et al. 2014;O’Gorman et al. 2015). The challenge remainsin determining whether these states are transi-

TF

Figure 1. CD4þ T-cell plasticity regulated by transcription factors (TF) that control the accessible chromatin.The process of differentiation lowers the potential of the cell to differentiate to other lineages as visualized byWaddington’s landscape. The cell eventually becomes committed to a lineage and resides in the valleys ofaccessible chromatin. New signaling events, through T-cell receptor (TCR) signaling and extracellular receptors,are propagated to the nucleus through the appropriate cytoplasmic machinery. This causes the expression oftranscription factors that modulate chromatin accessibility to allow for the cell to obtain new cell identity andfunction. Epigenetic regulatory mechanisms such as DNA methylation and histone modifications can eitherenhance or inhibit this process.






tory or stable and to determine their functionalstatus or disease relevance.

Integrating epigenetic measurements is anattractive solution for more precise definitionsof cell identity. The relative stability of cell iden-tity conferred by epigenetic mechanisms at thegenome-wide level makes this possible. Chro-matin accessibility measurements by DNase Ihypersensitivity or the new and robust assayATAC-seq has provided an excellent foundationof cell identity. A remarkable example is a recentcomparison of chromatin accessibility land-scapes in innate and adaptive T cells revealingsimilarity and differences in these two distinctcell types (Shih et al. 2016).

SINGLE-CELL HETEROGENEITY

Although not a novel observation, cellular het-erogeneity in once seemingly homogenouspopulations is even more apparent with the ad-vent of new technologies to take single-cell mea-surements. Variability in canonical subsets mayhelp fill gaps in proposed links between certainCD4 subsets and disease. Heterogeneity maynot be ultimately meaningful in most tissue sys-tems in multicellular organisms, as a certainlevel of homeostasis is required. However, theimmune system is one environment in whichcellular heterogeneity is of higher priority.Rare cells with the ability to respond to ever-changing pathogens provide the edge necessaryto overcome these pathogens. Indeed, heteroge-neity of the lymphocyte compartment encodedin the diverse antigen receptor repertoire hasgranted us a mechanism that ensures a smallproportion of the lymphocyte population willhave the means of recognition for nearly anynew pathogen. Much attention is being focusedon deconstructing the signaling and epigeneticmechanisms that give rise to variability and toassess the impact of that variability in the hostimmune response. As a result of recent geno-mics tools such as single-cell RNA-seq (Kolod-ziejczyk et al. 2015) and ATAC-seq (Buenrostroet al. 2015) together with revolutionary single-cell proteomics, we are poised to improve ourunderstanding of cell identity from a single-cellperspective.

LINEAGE TRACING IS IMPORTANTIN DETERMINING PAST, PRESENT,AND FUTURE

We posit that snapshot measurements of theCD4 compartment in response to an immunestimulus does not allow us to adequately resolvemultiple underlying populations at the epige-netic level. As discussed above, the influence ofprevious cytokine exposure can be recorded inthe chromatin. Understanding how this historymay influence future behavior and how a seem-ingly homogenous population of cells mightdiverge in response to new stimuli is criticallyimportant. If the goal is to understand the tran-scriptional circuitry that leads to polarized sub-sets of CD4 T cells, it stands to reason thatunderstanding how these epigenetics states cor-responds to function is of equal importance.Addressing these questions will not only requirean integration of genomics and phenotypicdata, but also better methods of lineage tracing.The advent of CRISPR-Cas9 brings the ability togenetically track cells in an ongoing manner andto serve as signposts from which we can take oursnapshot measurements of the cell to under-stand its trajectory (McKenna et al. 2016).

CONCLUDING REMARKS

Initiated by the application of microarrays, thelast decade unraveled the battery of transcrip-tion factors essential in T-helper-cell differenti-ation. Yet, we have limited knowledge about theefficacy of these proteins on the chromatin andtheir relative ability to alter the accessibilitylandscapes and ultimately T-cell fate in the pe-riphery. Novel techniques in genomics, epige-nomics, proteomics, single-cell biology, andgene editing together with computational tech-niques developed for data integration can helpus fill these gaps. Understanding the mecha-nisms underlying plasticity in CD4þ T cellswill provide us with the opportunity to tailorthe CD4þ responses for better vaccination andto reinvigorate T-cell responses in chronic in-fection and cancer. Overcoming the stable anddefined chromatin state through the expressionof transcription factors with effectuating prop-






erties can potentially overcome mechanisms ofinhibition and allows new signals—and fates—to be propagated to the nucleus to ultimatelycontrol the outcome of the immune response.

ACKNOWLEDGMENTS

This work is supported by the National Instituteof Allergy and Infectious Diseases (NIAID)(K22AI112570 to G.V). We thank Carol A.Clifton for Figure 1.

REFERENCES

Ahlfors H, Morrison PJ, Duarte JH, Li Y, Biro J, Tolaini M,Di Meglio P, Potocnik AJ, Stockinger B. 2014. IL-22 fatereporter reveals origin and control of IL-22 production inhomeostasis and infection. J Immunol 193: 4602–4613.

Allison KA, Sajti E, Collier JG, Gosselin D, Troutman TD,Stone EL, Hedrick SM, Glass CK. 2016. Affinity and doseof TCR engagement yield proportional enhancer andgene activity in CD4þ T cells. eLife 5: e10134.

Blau HM. 1989. How fixed is the differentiated state? Les-sons from heterokaryons. Trends Genet 5: 268–272.

Buenrostro JD, Wu B, Litzenburger UM, Ruff D, GonzalesML, Snyder MP, Chang HY, Greenleaf WJ. 2015. Single-cell chromatin accessibility reveals principles of regulato-ry variation. Nature 523: 486–490.

Ciofani M, Madar A, Galan C, Sellars M, Mace K, Pauli F,Agarwal A, Huang W, Parkurst CN, Muratet M, et al.2012. A validated regulatory network for th17 cell speci-fication. Cell 151: 289–303.

DuPage M, Bluestone JA. 2016. Harnessing the plasticity ofCD4þ T cells to treat immune-mediated disease. Nat RevImmunol 16: 149–163.

Garber M, Yosef N, Goren A, Raychowdhury R, Thielke A,Guttman M, Robinson J, Minie B, Chevrier N, Itzhaki Z,et al. 2012. A high-throughput chromatin immunopre-cipitation approach reveals principles of dynamic generegulation in mammals. Mol Cell 47: 810–822.

Gaudilliere B, Fragiadakis GK, Bruggner RV, Nicolau M,Finck R, Tingle M, Silva J, Ganio EA, Yeh CG, MaloneyWJ, et al. 2014. Clinical recovery from surgery correlateswith single-cell immune signatures. Sci Transl Med 6:255ra131.

Hirota K, Duarte JH, Veldhoen M, Hornsby E, Li Y, Cua DJ,Ahlfors H, Wilhelm C, Tolaini M, Menzel U, et al. 2011.Fate mapping of IL-17-producing T cells in inflammato-ry responses. Nat Immunol 12: 255–263.

Iwafuchi-Doi M, Zaret KS. 2016. Cell fate control by pioneertranscription factors. Development 143: 1833–1837.

Kolodziejczyk AA, Kim JK, Svensson V, Marioni JC, Teich-mann SA. 2015. The technology and biology of single-cellRNA sequencing. Mol Cell 58: 610–620.

Kurachi M, Barnitz RA, Yosef N, Odorizzi PM, DiIorio MA,Lemieux ME, Yates K, Godec J, Klatt MG, Regev A, et al.2014. The transcription factor BATF operates as an essen-

tial differentiation checkpoint in early effector CD8þ Tcells. Nat Immunol 15: 373–383.

McKenna A, Findlay GM, Gagnon JA, Horwitz MS, SchierAF, Shendure J. 2016. Whole-organism lineage tracing bycombinatorial and cumulative genome editing. Science353: aaf7907.

Mosmann TR, Coffman RL. 1989. TH1 and TH2 cells: Dif-ferent patterns of lymphokine secretion lead to differentfunctional properties. Annu Rev Immunol 7: 145–173.

Murphy KM, Stockinger B. 2010. Effector T cell plasticity:Flexibility in the face of changing circumstances. NatImmunol 11: 674–680.

Newell EW, Sigal N, Bendall SC, Nolan GP, Davis MM. 2012.Cytometry by time-of-flight shows combinatorial cyto-kine expression and virus-specific cell niches within acontinuum of CD8þ T cell phenotypes. Immunity 36:142–152.

O’Gorman WE, Hsieh EW, Savig ES, Gherardini PF, Her-nandez JD, Hansmann L, Balboni IM, Utz PJ, Bendall SC,Fantl WJ, et al. 2015. Single-cell systems-level analysis ofhuman Toll-like receptor activation defines a chemokinesignature in patients with systemic lupus erythematosus.J Allergy Clin Immunol 136: 1326–1336.

O’Shea JJ, Paul WE. 2010. Mechanisms underlying lineagecommitment and plasticity of helper CD4þ T cells. Sci-ence 327: 1098–1102.

Ostuni R, Piccolo V, Barozzi I, Polletti S, Termanini A, Bo-nifacio S, Curina A, Prosperini E, Ghisletti S, Natoli G.2013. Latent enhancers activated by stimulation in differ-entiated cells. Cell 152: 157–171.

Rivera CM, Ren B. 2013. Mapping human epigenomes. Cell155: 39–55.

Rothenberg EV. 2014. The chromatin landscape and tran-scription factors in T cell programming. Trends Immunol35: 195–204.

Shih HY, Sciume G, Mikami Y, Guo L, Sun HW, Brooks SR,Urban JFJr., Davis FP, Kanno Y, O’Shea JJ. 2016. Devel-opmental acquisition of regulomes underlies innate lym-phoid cell functionality. Cell 165: 1120–1133.

Soufi A, Donahue G, Zaret KS. 2012. Facilitators and im-pediments of the pluripotency reprogramming factors’initial engagement with the genome. Cell 151: 994–1004.

Stergachis AB, Neph S, Reynolds A, Humbert R, Miller B,Paige SL, Vernot B, Cheng JB, Thurman RE, SandstromR, et al. 2013. Developmental fate and cellular maturityencoded in human regulatory DNA landscapes. Cell 154:888–903.

Vahedi G, Takahashi H, Nakayamada S, Sun HW, SartorelliV, Kanno Y, O’Shea JJ. 2012. STATs shape the active en-hancer landscape of T cell populations. Cell 151: 981–993.

Vahedi G, Kanno Y, Furumoto Y, Jiang K, Parker SC, ErdosMR, Davis SR, Roychoudhuri R, Restifo NP, Gadina M, etal. 2015. Super-enhancers delineate disease-associatedregulatory nodes in T cells. Nature 520: 558–562.

van Oevelen C, Collombet S, Vicent G, Hoogenkamp M,Lepoivre C, Badeaux A, Bussmann L, Sardina JL, ThieffryD, Beato M, et al. 2015. C/EBPa activates pre-existingand de novo macrophage enhancers during induced pre-B cell transdifferentiation and myelopoiesis. Stem CellRep 5: 232–247.






Wilhelm C, Hirota K, Stieglitz B, Van Snick J, Tolaini M, LahlK, Sparwasser T, Helmby H, Stockinger B. 2011. An IL-9fate reporter demonstrates the induction of an innate IL-9 response in lung inflammation. Nat Immunol 12:1071–1077.

Wilson CB, Rowell E, Sekimata M. 2009. Epigenetic control ofT-helper-cell differentiation. Nat Rev Immunol 9: 91–105.

Yamane H, Paul WE. 2012. Memory CD4þ T cells: Fatedetermination, positive feedback and plasticity. CellMol Life Sci 69: 1577–1583.

Youngblood B, Oestreich KJ, Ha SJ, Duraiswamy J, AkondyRS, West EE, Wei Z, Lu P, Austin JW, Riley JL, et al. 2011.Chronic virus infection enforces demethylation of the

locus that encodes PD-1 in antigen-specific CD8þT cells.Immunity 35: 400–412.

Zaret KS, Mango SE. 2016. Pioneer transcription factors,chromatin dynamics, and cell fate control. Curr OpinGenet Dev 37: 76–81.

Zhou L, Chong MMW, Littman DR. 2009a. Plasticity ofCD4þ T cell lineage differentiation. Immunity 30: 646–655.

Zhou X, Bailey-Bucktrout SL, Jeker LT, Penaranda C, Mar-tınez-Llordella M, Ashby M, Nakayama M, Rosenthal W,Bluestone JA. 2009b. Instability of the transcription fac-tor Foxp3 leads to the generation of pathogenic memoryT cells in vivo. Nat Immunol 10: 1000–1007.






Do Memory CD4 T Cells Keep Their Cell-TypeProgramming: Plasticity versus FateCommitment?

T-Cell Heterogeneity, Plasticity, and Selectionin Humans

Federica Sallusto,1,2 Antonino Cassotta,1,2 Daniel Hoces,1 Mathilde Foglierini,1

and Antonio Lanzavecchia1

1Institute for Research in Biomedicine, Universita della Svizzera italiana, 6500 Bellinzona, Switzerland2Institute of Microbiology, ETH Zurich, 8093 Zurich, Switzerland


The wide range of effector and memory T cells is instrumental for immune regulation andtailored mechanisms of protection against pathogens. Here, we will focus on human CD4T cells and discuss T-cell plasticity and intraclonal diversification in the context of a progres-sive and selective model of CD4 T-cell differentiation.

OUTLINE

HETEROGENEITY OF HUMAN CD4 T CELLS, 10

PLASTICITY AND FATE COMMITMENT: HUMAN STUDIES, 12

T-CELL PLASTICITY AND INTRACLONAL DIVERSIFICATION, 13

REFERENCES, 15



9









CD4 T cells are born naıve and upon antigen-ic stimulation in secondary lymphoid

organs differentiate to effector T cells andacquire the capacity to migrate to peripheraltissue and to produce a range of cytokines thatmediate effector responses directly and throughthe activation of innate immune cells. TH1 cellsproduce interferon g (IFN-g) that activatesmacrophages to kill intracellular bacteria,whereas TH2 cells produce interleukin (IL)-4and IL-5, which activate mast cells, basophils,and eosinophils in response to helminthes, andTH17 cells produce IL-17 and IL-22, which driverecruitment of neutrophils and production ofantimicrobial peptides in response to extracel-lular bacteria and fungi. TH cells that migrate toB-cell follicles differentiate to follicular helper Tcells (TFH) and provide help to B cells for anti-body production. Following antigen-driven ex-pansion, some T cells survive as circulating cen-tral memory T cells (TCM) and effector memoryT cells (TEM) to provide immune surveillance inlymph nodes and peripheral tissues; others thatseed nonlymphoid tissues give rise to a popula-tion of tissue-resident memory T cells (TRM) toprovide immediate protection at sites of patho-gen reentry.

The heterogeneity of effector and memoryCD4 T cells raises questions about the natureof differentiation signals, the extent of fate di-versity, the lineage relationship between differ-ent fates, and the degree of plasticity at differentstages of differentiation. Here we will summa-rize studies performed in the human system thathave advanced our understanding of CD4 T-cellheterogeneity and discuss T-cell plasticity andintraclonal diversification in the context of aprogressive and selective model of CD4 T-celldifferentiation.

HETEROGENEITY OF HUMAN CD4 T CELLS

The extent and nature of naıve T-cell differen-tiation is determined by signals provided by thepathogen and by the local environment wherepriming occurs (Medzhitov 2007). By promot-ing expression of master transcription factors,cytokines produced by dendritic cells and otherinnate immune cells represent key determinants

of TH cell differentiation. Classical examples areIFN-g and IL-12, which induce through STAT1and STAT4 expression of T-bet that drives theTH1 program. Similarly, IL-4 through STAT6induces expression of GATA-3 that drives theTH2 program, while IL-6 through STAT3 inducesRORgt that drives the TH17 program. Impor-tantly, cytokines act in combination with signalsfrom the T-cell antigen receptor (TCR) andcostimulatory molecules as well as signals trig-gered by tissue-specific cues and environmentalfactors—such as nutrients, vitamins, and evenpollutants and salt—to define internal tran-scriptional landscapes that confer T-cell identi-ty (Bonelli et al. 2014). Some signals triggerinternal circuits that enforce, reinforce, ordestabilize T-cell polarization through multiplemechanisms, including epigenetic modifica-tions (Wang et al. 2015). MicroRNAs and longnoncoding RNAs (lncRNAs) have also beenfound to control various aspects of CD4 T-celldifferentiation and effector function (Baumjo-hann and Ansel 2013; Panzeri et al. 2015).

Mouse models have been used in a reduc-tionist approach to define the contribution ofsingle elements and signaling pathways involvedin CD4 T-cell differentiation and to evaluate therole of different types of polarized TH cells inprotection or immunopathology. Human stud-ies are providing important clues as to the ex-tent of T-cell fate diversity induced by differentpathogens in different tissues (Farber et al. 2014;Sallusto 2016). An example is represented bystudies analyzing the memory T-cell responseinduced by extracellular pathogens that elicittwo distinct types of TH17 cells. T cells primedby Staphylococcus aureus produce IL-17 and theimmune regulatory cytokine IL-10. In contrast,T cells primed by Candida albicans have a moreinflammatory phenotype, producing IL-17together with IFN-g. This difference was foundto be dependent on the cytokines elicited by thedifferent pathogens, primarily IL-1b (Zielinskiet al. 2012). Another example is provided byCD4 T cells in the skin of patients with atopicdermatitis or psoriasis that revealed the exis-tence of T cells producing IL-22 but not IL-17(Eyerich et al. 2009). These TH22 cells can bealso found in the blood, where they express

F. Sallusto et al.





CCR6 and the skin-homing receptors CLA andCCR10, and can differentiate from naıve T cellsin a process that requires the transcription fac-tor AHR and signals from vitamin D3 (Duhenet al. 2009; Trifari et al. 2009). Yet other exam-ples include IL-9-producing TH9 cells, initiallydefined in mice (Veldhoen et al. 2008) that arealso increased in the skin lesions of psoriasis(Schlapbach et al. 2014) and granulocyte mac-rophage colony-stimulating factor (GM-CSF)-only-producing TH cells (Noster et al. 2014).

Studies on the human T-cell response in-duced by Mycobacterium tuberculosis led to thediscovery of a distinct type of IFN-g-producingTH1 cells (that we defined as TH1�) that canbe distinguished from virus-induced TH1 cellsbased on the expression of chemokine receptorsand transcription factors (Acosta-Rodriguezet al. 2007; Lindestam Arlehamn et al. 2013).TH1� cells express CXCR3 and T-bet, as classicTH1 cells, but also CCR6 and RORgt, which arecharacteristic of TH17 cells (Fig. 1). Interesting-ly, TH1� development, but not TH1 develop-ment, is impaired in patients with RORC loss-

of-function mutations (Okada et al. 2015), afinding that supports the existence of distinctpathways of TH1 cell differentiation. The signalspresent in the context of bacterial infections thatinduce TH1� cells remain to be defined. Thesecells may derive directly from naıve T cells in aRORgt-dependent fashion or from TH17 cellsthat convert to TH1� under the influence ofIL-12, TNF-a, and/or IL-1b. The latter possi-bility is consistent with the finding that in TH1�

cells, like in TH17 cells, RORC2 and IL17A aredemethylated (Mazzoni et al. 2015).

A regulated cytokine production is requiredfor the proper elimination of microbial patho-gens, for instance IFN-g production by TH1cells for intracellular microbes and IL-17 byTH17 cells for C. albicans. However, when ana-lyzed in more detail, microbe-specific memoryT cells are found not only in the expected subsetbut also, at lower frequencies, in other subsetsthat have different, and even divergent, func-tions (Becattini et al. 2015). For instance,C. albicans–specific T cells are mainly TH17,but some are TH1, TH1�, or TH2. Similarly,

Bacteria(Mycobacterium tuberculosis)

Viruses(influenza virus)

Tn Tn

IL-12

IFN

IFN-γ IFN-γ

IL-12

IL-23

IL-1β

T-betRORγt

T-bet

CXCR3+

CCR6+CXCR3+

TH1 TH1*

Figure 1. Two types of human TH1 cells. CXCR3þCCR62 TH1 cells expressing T-bet are preferentially elicited byviruses, whereas CXCR3þCCR6þ TH1� cells expressing T-bet and RORgt are preferentially elicited by bacteria,possibly by different polarizing cytokines present at sites of induction. IFN, Interferon; IL, interleukin.






Mycobacteria-specific T cells are primarily TH1�,but a few are TH1 or TH17. The heterogeneityof T-cell fates revealed by human studies is notsurprising considering the continuous environ-mental exposure to commensal and pathogenicmicrobes and the accumulation of memoryT cells that have undergone repeated encounterswith antigen over several years. Although it ispossible that these heterogeneous fates are theresult of differential priming, it is also possiblethat they represent distinct stages of differenti-ation, owing to the property of T cells to acquireadditional functions or to be reprogrammedto alternative fates if exposed to appropriatestimuli, a property that has been defined asT-cell plasticity.

PLASTICITY AND FATE COMMITMENT:HUMAN STUDIES

Soon after the seminal discovery of TH1 andTH2 cells (Mosmann et al. 1986), some studiesboth in mice and humans reported the existenceof CD4 T-cell clones with a mixed cytokinesecretion pattern (Maggi et al. 1988; Umetsuet al. 1988; Street et al. 1990; Openshaw et al.1995). These TH0 clones producing IFN-g andIL-4 cells were suggested to be the precursorsof TH1 and TH2 cells (Street et al. 1990). Thisphenomenon was later analyzed by studyingepigenetic modifications at cytokine gene lociin naturally occurring human memory TH1 andTH2 cells (Messi et al. 2003). It was found thatmemory TH1 cells display acetylated histonesat the IFNG promoter, but not at the IL4 pro-moter, whereas reciprocally memory TH2 cellsdisplay acetylated histones at the IL4 but notIFNG promoter. However, the hypoacetylationof the nonexpressed cytokine gene did not leadto its irreversible silencing because, on stimula-tion under TH2 conditions, TH1 cells up-regu-lated GATA-3 and acquired IL4 acetylation andexpression while continuing to produce IFN-g,thus becoming TH0 cells. Reciprocally, whenstimulated in the presence of IL-12, most TH2cells up-regulated T-bet and acquired IFNGacetylation and expression while continuingto produce IL-4. These findings indicate thatmost in vivo–primed human TH1 and TH2 cells

maintain both memory and flexibility of cyto-kine gene expression.

Several studies have now provided convinc-ing evidence that most TH cells, in particularTH17 cells, have a great degree of flexibility.For instance, TH17 cells from the synovialfluid of oligoarticular-onset juvenile idiopathicarthritic patients can shift in vitro from theTH17 to the TH17/TH1 or TH1� phenotype(Cosmi et al. 2011). In patients with allergicasthma, TH17 cells also produce IL-4, suggest-ing a shift from TH17 to TH17/TH2 mixed phe-notype. Finally, IL-1b can induce conversion ofIL-10þ TH17 into more inflammatory IFN-gþ

TH17 cells (Zielinski et al. 2012). These exam-ples of flexibility in cytokine gene expressionunderline the robust and adaptive behavior ofeffector T cells in the immune response.

There is growing evidence that plasticity ismaximal at early stages of differentiation. Con-sistent with this notion, circulating CCR7þTCM

cells and CXCR5þ TFH-like cells, which repre-sent subsets of less differentiated T cells charac-terized by hypoacetylated cytokine genes, havethe potential to differentiate to either TH1 orTH2 when appropriately stimulated (Messi etal. 2003; Rivino et al. 2004). Plasticity in TFH

cells is consistent with the finding of subsetsof CXCR5þ T cells that have characteristicsof TH1, TH2, or TH17 cells (Morita et al.2011). Plasticity can be progressively lost as cellsreach terminal differentiation stages. For in-stance, TEM cells expressing CRTH2 and produc-ing high levels of IL-4 are unable to up-regulateT-bet and to acquire IFN-g-producing capacity(Messi et al. 2003). Although irreversible com-mitment appears to be the exception rather thanthe rule, these findings suggest that certainconditions of stimulations in vivo exist thatlead to irreversible commitment. These condi-tions may be found at special sites in the bodyor may develop in particular pathologic situa-tions. Sites of chronic inflammation may repre-sent antigen and cytokine rich environments inwhich effector T cells are continuously stimu-lated. Indeed, repeated stimulations of mouseCD4 T cells in the presence of IL-12 or IL-4enhance cytokine production and lead tostrongly polarized TH1 or TH2 cells (Murphy

F. Sallusto et al.





et al. 1996). It would be interesting to determinewhether reduced plasticity is a characteristic oftissue-resident memory T cells, which in hu-mans also represent a prominent subset of dif-ferentiated T cells (Thome and Farber 2015).Further studies are required to assess the plas-ticity of T cells at different stages of differentia-tion and in different anatomical compartments.

It is important to consider that plasticity isinfluenced by the expression of cytokine recep-tors and transcription factors in the respondingT cells and by the availability of polarizingsignals in the tissue where antigen reencounteroccurs. For instance, TH17 cells that expressIL-12R and IL-1R can acquire a TH1� phenotypein response to microbes that elicit production ofIL-12 or IL-1, a situation that may be found inthe course of coinfections. It can be speculatedthat, in the context of challenge by a relatedpathogen in a different tissue, the property ofplasticity may confer an evolutionary advantage.

T-CELL PLASTICITY AND INTRACLONALDIVERSIFICATION

The finding that the human T-cell response,even to a single microbial pathogen, is heteroge-neous could be explained by priming of multipleclonotypes, each undergoing a distinct differen-tiation pathway determined by the nature andstrength of antigen, costimulatory, and cytokinesignals. Alternatively, it is possible that withinan expanding T-cell clone the proliferatingcells might acquire different fates as the conse-quence of stochastic stimulation and plasticity.The “one cell–one fate” and “one cell–multiplefates” models were tested in humans by analyz-ing the distribution of clonotypes within mem-ory T-cell subsets (Becattini et al. 2015).

By combining antigenic stimulation andTCR deep sequencing on isolated subsets ofTH1, TH2, TH1�, and TH17 memory T cells, itwas shown that the same clonotype could befound in different subsets. C. albicans–specificclonotypes present in the TH17 subset were alsofound in the TH1� and TH2 subsets (Fig. 2A).This finding was strengthened by the isolationof T-cell clones with identical TCR ab, but dif-ferent surface marker and effector function, and

by in vitro priming experiments showing that asingle naıve T cell could generate multiple fatesin one round of stimulation with C. albicans(Becattini et al. 2015). The intraclonal differen-tiation toward multiple fates was even morestriking in the case of memory T cells inducedby the tetanus toxoid vaccine with several TCRVb sequences being found in TH1, TH2, TH1�,and TH17 subsets (Fig. 2B).

Given the overwhelming evidence for intra-clonal diversification, how can we explain high-ly polarized responses to pathogens? Interest-ingly, the dominant TH17 response inducedin vivo by C. albicans was because the cloneswithin this subset were more expanded, ratherthan more numerous, as compared to the clonesin the other subsets. This finding would be con-sistent with the initial priming of multiple typesof T cells, followed by the selective expansion ofTH17 cells, which may be determined by theprevailing cytokines at the site of priming orin the tissue where the pathogen is encountered.Interestingly, this was not the case for T cellsprimed by a weak vaccine such as tetanus toxoidthat failed to select for a particular type of po-larized cells. The generation of a range of effec-tor and memory T cells within the same clonemay be consistent with an early diversificationor with the priming of uncommitted CD4 T-cellclones that subsequently progress along multi-ple differentiation pathways (Lanzavecchia andSallusto 2000).

In the context of heterogeneous T-cell re-sponses, an important question that needs tobe addressed is whether the generation of func-tionally diverse T-cell subsets is advantageousfor the host. Although failure to generate theright class of TH cell response may lead tosusceptibility to infections, it is possible thatthe induction of a spectrum of effector andmemory T cells endowed with different migra-tory capacities would provide the host with arange of differentiated precursors to be recruit-ed and expanded where necessary.

Moving ahead, the challenge is to dissect interms of time and space the events that deter-mine the functional heterogeneity of T-cellresponses in humans. Improved technologiesfor obtaining information on antigen specific-






TH17

T H1

TH2

T H1*

Tet

anus

toxo

id

TH17

T H1

TH2T H

1*

Can

dida

alb

ican

sA

B

140

120

100

80

60

40

20

0

180

160

140

120

100

80

60

40

20

20

0

160

180

200

220

240

260

280

300

0

20

40

60

80

100

120

140

160

180

200

220

240

260

280

320

300

340

360

380

400

420

0

2040608010012014

016018

0

200

220

240

260

280

300

320

340

360

380

0

40

60

80

100

120140

160180

200220

24026

0

280 30

0 320 34

0 360 38

0 400 42

0 440 46

0

480

0 20 40 60 80 100

120

140

160

180

200

220

240

260

280

300

320

340

360

380

400

420

440

460

480

500

520

540

560

580

600

620

640

0

20

40

60

80

100

120

140

160

180

200

220

240

260

280

300

320

340

360

380

400

420

0

20406080100

120

140

16018020

022024

026028

030032

034036

0380

400420

440460

480500

520

540

560

580

600

Figu

re2.

Gra

ph

ical

rep

rese

nta

tio

no

fth

eex

ten

tofs

har

ing

amo

ng

T-c

ella

nti

gen

rece

pto

r(T

CR

)cl

on

oty

pes

ofm

emo

ryT

H1

(red

),T

H2

(yel

low

),T

H1�

(blu

e),a

nd

TH

17(g

reen

)ce

lls

pri

med

by

Can

did

aal

bica

ns

(A),

or

teta

nu

sto

xoid

(B).

Ou

ter

rin

gsan

dco

nn

ecti

ng

lin

esid

enti

fycl

on

oty

pes

shar

edb

etw

een

two

or

mo

reT

-cel

lsu

bse

ts.

Dat

aar

ere

pre

sen

tati

veo

fsi

ngl

ed

on

ors

.(F

rom

Bec

atti

ni

etal

.20

15;

mo

difi

ed,

wit

hp

erm

issi

on

.)

F. Sallusto et al.





ity, TCR sequence, and phenotype from singleT cells will likely unravel even further degreesof interclonal and intraclonal heterogeneityin the T-cell response to microbial pathogensand vaccines (Han et al. 2014). Resolving thebehavior of T-cell populations at the level ofindividual cells and in the context of protectiveor pathological immune responses is a majorchallenge that holds promise for a better under-standing of the immune system and for im-mune interventions.

REFERENCES

Acosta-Rodriguez EV, Rivino L, Geginat J, Jarrossay D, Gat-torno M, Lanzavecchia A, Sallusto F, Napolitani G. 2007.Surface phenotype and antigenic specificity of humaninterleukin 17-producing T helper memory cells. NatImmunol 8: 639–646.

Baumjohann D, Ansel KM. 2013. MicroRNA-mediatedregulation of T helper cell differentiation and plasticity.Nat Rev Immunol 13: 666–678.

Becattini S, Latorre D, Mele F, Foglierini M, De Gregorio C,Cassotta A, Fernandez B, Kelderman S, Schumacher TN,Corti D, et al. 2015. T cell immunity. Functional hetero-geneity of human memory CD4þ T cell clones primed bypathogens or vaccines. Science 347: 400–406.

Bonelli M, Shih HY, Hirahara K, Singelton K, Laurence A,Poholek A, Hand T, Mikami Y, Vahedi G, Kanno Y, et al.2014. Helper T cell plasticity: Impact of extrinsic andintrinsic signals on transcriptomes and epigenomes.Curr Top Microbiol Immunol 381: 279–326.

Cosmi L, Cimaz R, Maggi L, Santarlasci V, Capone M, Bor-riello F, Frosali F, Querci V, Simonini G, Barra G, et al.2011. Evidence of the transient nature of the Th17 phe-notype of CD4þCD161þ T cells in the synovial fluidof patients with juvenile idiopathic arthritis. ArthritisRheum 63: 2504–2515.

Duhen T, Geiger R, Jarrossay D, Lanzavecchia A, Sallusto F.2009. Production of interleukin 22 but not interleukin 17by a subset of human skin-homing memory T cells. NatImmunol 10: 857–863.

Eyerich S, Eyerich K, Pennino D, Carbone T, Nasorri F, Pal-lotta S, Cianfarani F, Odorisio T, Traidl-Hoffmann C,Behrendt H, et al. 2009. Th22 cells represent a distincthuman T cell subset involved in epidermal immunity andremodeling. J Clin Invest 119: 3573–3585.

Farber DL, Yudanin NA, Restifo NP. 2014. Human memoryT cells: Generation, compartmentalization and homeo-stasis. Nat Rev Immunol 14: 24–35.

Han A, Glanville J, Hansmann L, Davis MM. 2014. LinkingT-cell receptor sequence to functional phenotype at thesingle-cell level. Nat Biotechnol 32: 684–692.

Lanzavecchia A, Sallusto F. 2000. Dynamics of T lymphocyteresponses: Intermediates, effectors, and memory cells.Science 290: 92–97.

Lindestam Arlehamn CS, Gerasimova A, Mele F, HendersonR, Swann J, Greenbaum JA, Kim Y, Sidney J, James EA,

Taplitz R, et al. 2013. Memory T cells in latentMycobacterium tuberculosis infection are directed againstthree antigenic islands and largely contained in aCXCR3þCCR6þ Th1 subset. PLoS Pathog 9: e1003130.

Maggi E, Del Prete G, Macchia D, Parronchi P, Tiri A, Chre-tien I, Ricci M, Romagnani S. 1988. Profiles of lympho-kine activities and helper function for IgE in human T cellclones. Eur J Immunol 18: 1045–1050.

Mazzoni A, Santarlasci V, Maggi L, Capone M, Rossi MC,Querci V, De Palma R, Chang HD, Thiel A, Cimaz R, et al.2015. Demethylation of the RORC2 and IL17A in humanCD4þ T lymphocytes defines Th17 origin of nonclassicTh1 cells. J Immunol 194: 3116–3126.

Medzhitov R. 2007. Recognition of microorganisms andactivation of the immune response. Nature 449: 819–826.

Messi M, Giacchetto I, Nagata K, Lanzavecchia A, Natoli G,Sallusto F. 2003. Memory and flexibility of cytokine geneexpression as separable properties of human TH1 andTH2 lymphocytes. Nat Immunol 4: 78–86.

Morita R, Schmitt N, Bentebibel SE, Ranganathan R, Bour-dery L, Zurawski G, Foucat E, Dullaers M, Oh S, Sabz-ghabaei N, et al. 2011. Human blood CXCR5þCD4þ Tcells are counterparts of T follicular cells and containspecific subsets that differentially support antibody secre-tion. Immunity 34: 108–121.

Mosmann TR, Cherwinski H, Bond MW, Giedlin MA,Coffman RL. 1986. Two types of murine helper T cell clone.I: Definition according to profiles of lymphokine activitiesand secreted proteins. J Immunol 136: 2348–2357.

Murphy E, Shibuya K, Hosken N, Openshaw P, Maino V,Davis K, Murphy K, O’Garra A. 1996. Reversibility of Thelper 1 and 2 populations is lost after long-term stimu-lation. J Exp Med 183: 901–913.

Noster R, Riedel R, Mashreghi MF, Radbruch H, Harms L,Haftmann C, Chang HD, Radbruch A, Zielinski CE.2014. IL-17 and GM-CSF expression are antagonisticallyregulated by human T helper cells. Sci Transl Med 6:241ra280.

Okada S, Markle JG, Deenick EK, Mele F, Averbuch D, LagosM, Alzahrani M, Al-Muhsen S, Halwani R, Ma CS, et al.2015. Impairment of immunity to Candida and Myco-bacterium in humans with bi-allelic RORC mutations.Science 349: 606–613.

Openshaw P, Murphy EE, Hosken NA, Maino V, Davis K,Murphy K, O’Garra A. 1995. Heterogeneity of intracel-lular cytokine synthesis at the single-cell level in polarizedT helper 1 and T helper 2 populations. J Exp Med 182:1357–1367.

Panzeri I, Rossetti G, Abrignani S, Pagani M. 2015. Longintergenic non-coding RNAs: Novel drivers of humanlymphocyte differentiation. Front Immunol 6: 175.

Rivino L, Messi M, Jarrossay D, Lanzavecchia A, Sallusto F,Geginat J. 2004. Chemokine receptor expression identi-fies Pre-T helper (Th)1, Pre-Th2, and nonpolarized cellsamong human CD4þ central memory T cells. J Exp Med200: 725–735.

Sallusto F. 2016. Heterogeneity of human CD4þ T cellsagainst microbes. Annu Rev Immunol 34: 317–334.

Schlapbach C, Gehad A, Yang C, Watanabe R, Guenova E,Teague JE, Campbell L, Yawalkar N, Kupper TS, Clark RA.2014. Human TH9 cells are skin-tropic and have auto-






crine and paracrine proinflammatory capacity. Sci TranslMed 6: 219ra218.

Street NE, Schumacher JH, Fong TA, Bass H, Fiorentino DF,Leverah JA, Mosmann TR. 1990. Heterogeneity of mousehelper T cells. Evidence from bulk cultures and limitingdilution cloning for precursors of Th1 and Th2 cells.J Immunol 144: 1629–1639.

Thome JJ, Farber DL. 2015. Emerging concepts in tissue-resident T cells: Lessons from humans. Trends Immunol36: 428–435.

Trifari S, Kaplan CD, Tran EH, Crellin NK, Spits H. 2009.Identification of a human helper T cell populationthat has abundant production of interleukin 22 and isdistinct from TH-17, TH1 and TH2 cells. Nat Immunol 10:864–871.

Umetsu DT, Jabara HH, DeKruyff RH, Abbas AK, AbramsJS, Geha RS. 1988. Functional heterogeneity amonghuman inducer T cell clones. J Immunol 140: 4211–4216.

Veldhoen M, Uyttenhove C, van Snick J, Helmby H, West-endorf A, Buer J, Martin B, Wilhelm C, Stockinger B.2008. Transforming growth factor-b “reprograms” thedifferentiation of T helper 2 cells and promotes an inter-leukin 9-producing subset. Nat Immunol 9: 1341–1346.

Wang C, Collins M, Kuchroo VK. 2015. Effector T cell dif-ferentiation: Are master regulators of effector T cells stillthe masters? Curr Opin Immunol 37: 6–10.

Zielinski CE, Mele F, Aschenbrenner D, Jarrossay D, Ronchi F,Gattorno M, Monticelli S, Lanzavecchia A, Sallusto F. 2012.Pathogen-induced human TH17 cells produce IFN-g orIL-10 and are regulated by IL-1b. Nature 484: 514–518.

F. Sallusto et al.





Do Memory CD4 T Cells Keep Their Cell-TypeProgramming: Plasticity versus FateCommitment?Complexities of Interpretation due to theHeterogeneity of Memory CD4 T Cells,Including T Follicular Helper CellsShane Crotty1,2,3

1Division of Vaccine Discovery, La Jolla Institute for Allergy and Immunology, La Jolla, California 920372Scripps Center for HIV/AIDS Vaccine Immunology and Immunogen Discovery (CHAVI-ID), La Jolla,California 92037

3Department of Medicine, Division of Infectious Diseases, University of California, San Diego, La Jolla,California 92093


Plasticity is the ability of a cell type to convert to another cell type. There are multiple effectorCD4 T-cell subtypes, including TH1, TH2, TH17, TH1�, CD4 CTL, TH9, and TFH cells. It iscommonly thought that a CD4 T cell can readily show full plasticity—full conversion fromone differentiated cell—and this propensity to plasticity is possessed by memory CD4 T cells.However, there remains no direct demonstration of in vivo–generated resting memory CD4T-cell conversion to a different subtype on secondary antigen challenge in vivo in an intactanimal at the single-cell level. What has been clearly shown is that CD4 T cells possessextraordinary capacity for phenotypic heterogeneity, but that is a distinct property fromplasticity. Heterogeneity is diversity of the resting memory CD4 T-cell population, notconversion of a single differentiated cell into another subtype. Apparently, plasticity at thepopulation level can be accomplished by either mechanism, as heterogeneity of CD4 T-cellsubpopulations could affect large shifts in subtype distribution at the overall population levelvia differential exponential expansion and death.

OUTLINE

STABILITY DURING A PRIMARY RESPONSE, 18STABILITY DURING TRANSITION FROM EFFECTOR CELL TO MEMORY CELL, 19MEMORY CD4 T-CELL PHENOTYPE HETEROGENEITY AND STABILITY DURING

RESTING MEMORY, 22PLASTICITY WHEN MEMORY CD4 T CELLS ARE RECALLED?, 23CONCLUDING REMARKS, 24ACKNOWLEDGMENTS, 25REFERENCES, 25



17







Here, plasticity is defined as the conversionof a single cell possessing a well-character-

ized CD4 T-cell type into a cell no longer pos-sessing that phenotype and instead possessing adifferent well-characterized CD4 T-cell pheno-type. For example, conversion of a memory TH1cell (T-betþIFN-gþCXCR52Bcl62) into a TFH

cell (Bcl6þCXCR5þT-bet2IFN-g2) would beplasticity. Separately, heterogeneity within awell-characterized CD4 T-cell population is de-fined here as a collection of varied phenotypes(,100% of the cell population) linked by ashared core phenotype. For example, heteroge-neity among TH1 cells can be observed by flowcytometry or mass cytometry by defining TH1cells as T-betþIFN-gþ cells and then observingfractions of the population expressing tumornecrosis factor (TNF), or interleukin (IL)-2,or Blimp1, or IL-10, or Eomes, etc. As anotherexample, heterogeneity among TH2 cells can beobserved by flow cytometry or mass cytometryby defining TH2 cells as GATA3hi cells and thenobserving fractions of the population express-ing IL-5, IL-4, IL-13, CRTH2, CCR4, or IL-10,etc. As another example, heterogeneity amonggerminal center (GC) TFH cells can be observedby flow cytometry or mass cytometry by defin-ing GC TFH cells as Bcl6þCXCR5þ cells andthen observing fractions of the population ex-pressing CXCL13, IL-21, IL-4, or CXCR3, etc.(Crotty 2014; Vinuesa et al. 2016). Heterogene-ity at the whole population level further includesthe range of differentiated CD4 T-cell subtypespresent, including TH1, TH2, TH17, TH1�, CD4CTL, TH9, and TFH cells, and perhaps evensome form of “unbiased” TH0-type cells. Bothplasticity and heterogeneity must be describedbased on analyses at the single-cell level.

Reports of T-cell program plasticity are un-convincing when the data are population-levelchanges in phenotypes. Such results can easilybe the outcome of outgrowth of a minor cellpopulation to become the dominant cell popu-lation, or vice versa, particularly given the ex-ponential proliferation that T cells are capableof. Also unconvincing are the relevance ofreports of cell program plasticity for whichthe central experiments are cell transfers intonew hosts, particularly new hosts that are

severely immunocompromised (e.g., T-cell-deficient mice). Such experiments show thatCD4 T-cell plasticity can occur under extremeconditions, but the experiments have no dem-onstrated relevance to what CD4 T cells actuallydo or experience in an intact animal. In con-trast, if transferred cells do maintain stability,those results are more credible, because theyshow stability of cell identity even when exposedto nonphysiological stresses. Apparent plasticityof differentiated CD4 T cells in vitro is generallynot convincing, both because the in vitro exper-iments lack demonstrated in vivo relevanceand because the experiments are performedat the cell population level, masking the impactof outgrowth of minor cell populations. Thestrictest criterion for demonstration of plastic-ity is the use of a lineage marker reporter trans-genic mouse, tracking, over time, cells markedirreversibly. Such an experiment directly estab-lishes the transcriptional history of a given cell.Many lineage-tracking experiments have beenperformed on nTregs, making use of Foxp3-IRES-GFP/YFP/RFP-Cre-based designs (Rub-tsov et al. 2008, 2010; Zhou et al. 2009; Miyaoet al. 2012). The central conclusions from thetwo later studies with more sophisticated mod-ified Foxp3 gene reporter constructs was thatFoxp3þ nTregs are very stable, with almost noplasticity (Rubtsov et al. 2010; Miyao et al.2012). In contrast, substantial gene-expressionheterogeneity could be observed in conditionsof stress and while still maintaining core Foxp3þ

nTreg programming. Still, the stability conclu-sions drawn from such studies are not necessar-ily directly transferrable for antigen-specificCD4 T-cell responses and CD4 T-cell memory,because nTregs develop their initial program-ming during thymic development.

STABILITY DURING A PRIMARY RESPONSE

There are no lineage marker reporter mousestudies showing plasticity of TH1, TH2, TH17,or TFH cells during a primary immune responsein an intact animal. Thus, excluding thymic-derived Tregs, there is no definitive evidenceof physiologically relevant CD4 T-cell plasticityduring a primary immune response. Cell-trans-

S. Crotty





fer experiments have attempted to addressstability or plasticity of antigen-specific CD4 Tcells during a primary immune response. Weobserved that TFH and TH1 cells during a viralinfection establish largely irreversible cell fatesby 72 h postinfection, based on cell transfersof virus-specific TH1 or TFH cells from virallyinfected mice into time-matched virally infect-ed mice (Choi et al. 2013). Similar pronouncedcell-fate commitment results were indepen-dently reported using a protein immunizationand an RFP-Bcl6 reporter mouse strain whentransferring CXCR52Bcl62 or CXCR5þBcl6þ

cells at day 7 postinfection (Liu et al. 2012).Plasticity of TH1 and TH2 cells to become TFH

cells has been reported; however, those experi-ments used in vitro–generated TH1 and TH2cells transferred into mice (Liu et al. 2012) orin vitro polarized cells then repolarized underdifferent in vitro conditions (Lu et al. 2011). It isalmost certainly the case that there is a windowof time early during effector CD4 T-cell differ-entiation in a primary immune response when agiven CD4 T cell possesses pluripotency, simul-taneously expresses lineage-defining transcrip-tion factors (e.g., Bcl6 and T-bet and RORgt)(Nakayamada et al. 2011; Oestreich et al. 2012),and maintains the capacity to respond to differ-ent extrinsic signals and subsequently committo one differentiated cell type (e.g., TFH or TH1or TH17) (DuPage and Bluestone 2016). Thus,simple questions regarding durable stabilityversus plasticity must be assessed after thatpoint, which is nontrivial to accomplish.

STABILITY DURING TRANSITION FROMEFFECTOR CELL TO MEMORY CELL

The transition from an effector CD4 T cell to acentral memory CD4 T cell appears to also bea transition from a cell with a highly polarizedgene-expression program to a cell with a lesspolarized gene-expression program. This maybe key to understanding the apparent plasticityof memory CD4 T cells, discussed below.

Based on single-cell transfer studies inmouse model systems, most CD4 T-cell clonesare capable of generating memory cells (Tuboet al. 2016), and a given individual CD4 T-cell

clone can differentiate into multiple differentCD4 T-cell types (e.g., TFH and TH1) as theydivide during a primary immune response(Tubo et al. 2013). Furthermore, those effectorcells can then develop into memory TFH and TH1cells in frequencies comparable with the fre-quencies of TFH and TH1 cells generated bythat clone during the effector phase of the CD4T-cell response (Tubo et al. 2016). Human T-cellreceptor (TCR) sequencing clonotype analysis ofantigen-specific human memory CD4 T cells hasshown that a given TCR sequence can be foundin TH1, TH2, and TH17 antigen-specific centralmemory cells (Becattini et al. 2015), consistentwith the mouse model observation.

During a primary immune response, ithas been observed that TFH cells can have geneexpression of other T-cell differentiation pro-grams. In the context of a mouse with an acutelymphocytic choriomeningitis virus (LCMV)infection, the mantle TFH cells (mTFH, outsideof GCs) and GC TFH cells express T-bet andinterferon g (IFN-g) at substantially higheramounts than naıve CD4 T cells (Johnstonet al. 2009; Yusuf et al. 2010; Ray et al. 2015).In our first paper on TFH cells, we stated “it isnotable that T-bet and IFN-g were still ex-pressed in the TFH in vivo, although at lowerlevels than in TH1/non-TFH LCMV-specificCD4 T cells. These observations are consistentwith a model in which TFH cells follow their owndifferentiation pathway but are not an isolatedlineage and can show partial characteristics ofTH1/TH2 polarization depending on environ-mental conditions.” Similar observations havebeen made for simian immunodeficiency virus(SIV) infection of rhesus macaques (Iyer et al.2015). Given that both LCMV and SIV infec-tions are extreme TH1-biased immune respons-es, the presence of TH1 gene expression by TFH

cells in LCMV, and SIV immune response mayrepresent uncommon exceptions. In supportof that concept, human tonsillar GC TFH cellsexpressing TH1, TH2, or TH17 cytokines arerarely observed (Ma et al. 2009; Yu et al. 2009;Dan et al. 2016; Havenar-Daughton et al. 2016).For example, ,1% of GC TFH cells produce IL-17 (Yu et al. 2009; Wong et al. 2015; Dan et al.2016). Mouse GC TFH cells rarely produce IL-






13, even under very strong TH2 polarizing hel-minth infection conditions (Liang et al. 2012),or house dust mite sensitization (Ballesteros-Tato et al. 2016) (TFH cells normally expressIL-4 as part of canonical TFH programming,distinct and independent of TH2 programming,and thus TFH expression of IL-4 is not an indi-cation of any TH2 gene programming [Crotty2011]). Furthermore, Bcl6 represses many TH1,TH2, and TH17 genes and can prevent TH1 dif-ferentiation (Johnston et al. 2009; Oestreichet al. 2012; Hatzi et al. 2015). A counterargu-ment can now be made that cytokine expressionby intracellular cytokine staining is insufficient-ly sensitive to determine whether a GC TFH cellmay possess TH1, or TH2, or TH17 gene expres-sion, because GC TFH cells are intrinsicallystingy cytokine producers, and intracellularcytokine staining missed �98% of human ormacaque antigen-specific GC TFH cells (Danet al. 2016; Havenar-Daughton et al. 2016).Thus, single-cell RNAseq of GC TFH cells maybe required to better understand whether GCTFH cells with partial TH1, TH2, or TH17 het-erogeneity characteristics are common or rare.

Considering the process from the oppositedirection, it is clear that human memory TFH

cells can show certain phenotypic markers com-monly associated with TH1, TH2, or TH17 cells(discussed more in the next section). What isthe ontogeny of those cells? One possibility isthat TFH cells can be imprinted with a fractionalamount of TH1 gene programming by an anti-gen-presenting cell during the early stages ofT-cell priming in response to a TH1 pathogen,and, although that gene-expression programis efficiently squelched by Bcl6 in the effectormTFH and GC TFH progeny of that cell, a partialTH1 program remains imprinted (Fig. 1). As aGC TFH cell transitions into a memory cell, itloses expression of Bcl6 protein (Kitano et al.2011; Liu et al. 2012; Choi et al. 2013; Hale et al.2013; Locci et al. 2013; Ise et al. 2014), thusderepressing a range of genes, including Ccr7and Il7ra (Fig. 1) (Kitano et al. 2011; Choiet al. 2013). One possibility is that the loss ofBcl6 protein as the GC TFH cell transitions into amemory TFH cell can allow a partial TH1 pro-gram imprinted at the time of T-cell priming to

then become derepressed in some cells in theabsence of Bcl6 protein, resulting in the centralmemory TFH cell acquiring a partially mixedTFH/TH1 phenotype (Fig. 1, model 1). Suchan event may be considered “imprinted partialplasticity” because the phenotype of the cellwould be dependent on the initial signals it re-ceived during T-cell priming, even if the gene-expression program was kept silent for longperiods of time. If the cell were to subsequentlybecome a TH1 cell and lose TFH characteristics,that would be “imprinted full plasticity.” Thisconcept has not been directly tested.

A second possibility (model 2) is that a GCTFH cell that does not have any TH1, TH2, orTH17 gene expression or imprinting may be in-duced to activate a partial TH1, TH2, or TH17gene-expression program if there are TH1, TH2,or TH17 differentiation inductive signals stillpresent in the environment when the GC TFH

cell is transitioning to become a memory celland losing Bcl6 protein expression (Fig. 1).Such an event would be CD4 T-cell plasticity,termed “de novo partial plasticity” here, to dis-tinguish it from imprinted plasticity.

Alternatively, a substantial proportion ofmemory TFH cells may be generated very earlyduring an immune response derived from man-tle TFH cells (mTFH) without going through aGC TFH cell stage. Evidence for such a pathwaycomes from studies of Sh2d1a2/2 mice andhumans, which have CD4 T cells that can differ-entiate into mTFH cells but not GC TFH cells andhave evidence of TFH cell memory (He et al.2013). If memory TFH cells are generated viasuch a pathway in immunocompetent miceand humans, it is plausible that memory TFH

cells derived from mTFH cells may be less polar-ized than memory TFH cells derived from GCTFH cells, with less fixed-fate programming, andtherefore may be more likely to have mixed at-tributes of TFH and TH1 or other programs(model 3). As is the case for models 1 and 2,this concept has also not been directly tested.

Given the observations and models de-scribed above, the presence of CXCR3þ

CXCR5þ memory CD4 T cells in humanperipheral blood is consistent with differentia-tion models based on heterogeneity, imprinted

S. Crotty





Model 2. De novo partial plasticity

= TH1 bias

DC

DC + TH1stimuli

mTFH GC TFH TransitioningTFH

MemoryTFH

Model 1. Imprinted partial plasticity

2° TFH

+ Ag

CXCR5+Bcl6+

IFN-γ+CXCR3+T-bet+mTFH1 GC TFH TransitioningTFH1

MemoryTFH1

2° TFH

+ Ag2° TFH1

2° TH1

CXCR5+PD-1+Bcl6hi

IFN-γ loCXCR3–T-betloCXCR5hiPD-1hiBcl6hi

IFN-γ–CXCR3–T-betloCXCR5+PD-1+Bcl6lo

IFN-γ–CXCR3–T-betloCXCR5+PD-1–/loBcl6lo

IFN-γ–CXCR3–T-betloCXCR5+PD-1+Bcl6hi

IFN-γ–CXCR3–T-betlo

CXCR5+PD-1+Bcl6hi

IFN-γ+CXCR3+T-bet+CXCR5hiPD-1hiBcl6hi

IFN-γ–CXCR3–T-bet–CXCR5+PD-1+Bcl6lo

IFN-γ+CXCR3+T-bet+CXCR5+PD-1–/loBcl6lo

IFN-γ+CXCR3+T-betlo

CXCR5–Bcl6–

IFN-γ+CXCR3+T-bethi

CXCR5+Bcl6+

IFN-γ–CXCR3–T-bet–

CXCR5+Bcl6+

IFN-γ+CXCR3+T-bet+

2° TFH

2° TFH1

2° TH1

2° TFH

2° TFH1

2° TH1

CXCR5–Bcl6–

IFN-γ+CXCR3+T-bethi

CXCR5+Bcl6+


mTFH GC TFH

Transitioning TFH

TH1inflammatoryenvironment

MemoryTFH 2° TFH

+ Ag

TransitioningTFH1

MemoryTFH1

CXCR5+Bcl6+


CXCR5+PD-1+Bcl6lo

IFN-γ+CXCR3+T-bet+

Model 3. Early generated memory cells have greater plasticity

mTFH GC TFH

Transitioning TFH

MemoryTFH

?

?

?

?

Transitioning TFH

MemoryTFH 2° TFH

+ Ag

+ Ag

+ Ag

?

?

Figure 1. Three models of the development of heterogeneous or plastic CD4 T-cell memory. Each model isdiscussed in the main text.






partial plasticity, or de novo partial plasticity.CXCR3 is expressed by TH1 cells and is a directtarget of T-bet, and the CXCR3þ CXCR5þ

memory CD4 T cells in human peripheralblood cells are almost all capable of producingIFN-g on stimulation (Morita et al. 2011;Bentebibel et al. 2013; Locci et al. 2013;Obeng-Adjei et al. 2015). Thus, CXCR3þ

CXCR5þ memory CD4 T cells could havepotentially derived from CXCR3þ GC TFH cellsor mTFH cells (Iyer et al. 2015), or CXCR32 GCTFH cells or mTFH cells that were exposed toa TH1 environment while transitioning tomemory TFH cells.

CD4 T-cell biology usually does not fit tidysingle pathway models; heterogeneity of pheno-types and differentiation patterns are common,as this is likely important to confound pathogenimmune evasion strategies (Crotty 2012). Thus,a new report is surprising, but perhaps shouldnot have been. Development of TH2 cells (IL-5þ

IL-13þ CXCR52) in the lungs in response to asecond exposure to house dust mite antigenswas observed to be dependent on effector TFH

cells, with multiple lines of evidence pointing todifferentiation of GC TFH cells (CXCR5þPD-1hi

IL-21þ) into TH2 cells (Ballesteros-Tato et al.2016). These appeared to be fully differentiatedactive GC TFH cells, to the best of the ability ofthe authors to sort a pure cell population, withthe previously stated caveats. In contrast, adifferent group, using a different IL-21 reportermouse, did not observe TFH cells to be precur-sors to TH2 cells (IL-33Rþ IL-5þ IL-13þ) in asimilar house dust mite model (Coquet et al.2015), but they did not gate on CXCR5þ cellsfor the cell sorts. The two groups also trans-ferred cells at different times after the primaryantigen exposure, which may result in trackingcells at different points of cell-fate commitment.When TFH!TH2 plasticity occurred, the TFH

cells were taken 6 d after the primary immuni-zations (Ballesteros-Tato et al. 2016). In neitherstudy were resting memory CD4 T cells used(cells without activation marker expressiontaken at .30 d after the last antigen exposure).Lineage-tracking models that do not depend oncell transfers are likely to be the only means ofresolving such disparate observations.

MEMORY CD4 T-CELL PHENOTYPEHETEROGENEITY AND STABILITY DURINGRESTING MEMORY

Resting memory CD4 T cells appear to be largelystable over time by major lineage-defining phe-notypic markers, as shown for antigen-specificTFH, TH1, and TH2 cells in mouse models(Harrington et al. 2008; Hale et al. 2013;Hondowicz et al. 2016; Tubo et al. 2016). Humandata support the same conclusion but the anti-gen-specific data are limited (Locci et al. 2013;Bancroft et al. 2016; Da Silva Antunes et al.2017). No longitudinal data on antigen-specificresting memory CD4 T-cell phenotypes from in-dividual human donors are available at single-cell flow-cytometric resolution, which is muchneeded to demonstrate memory CD4 T-cell sub-set stability.

TH17 cells may be an exception to memory.There remains little evidence showing cleardemonstration of in vivo–generated TH17memory cells in mice. TH17 memory was absentin intact mice in one longitudinal antigen-spe-cific CD4 T-cell study (Pepper et al. 2010). Alater paper reported TH17 memory, but it de-pended on transfer of in vitro–generated TH17cells (Muranski et al. 2011). Although an Il17alineage marking reporter mouse has beenavailable for many years (Hirota et al. 2011),there is a lack of publications on in vivo–gen-erated TH17 memory. In humans, presence ofantigen-experienced TH17 cells to Candida al-bicans has been clearly demonstrated (Zielinskiet al. 2012); however, recurrent or continualexposure was not excluded, and a resting mem-ory TH17 phenotype (e.g., Ki672) was notshown for C. albicans–specific cells. Thus,although the existence of resting stable memoryTH17 cells seems biologically reasonable andthere is indirectly supportive literature (Linden-strøm et al. 2012; McGeachy 2013), data show-ing antigen-specific resting memory TH17 cellswith single-cell analysis are currently quitelimited.

Although there has been evidence of hetero-geneity within CD4 T-cell subsets, going back toearly descriptions of TH1 and TH2 cells (e.g.,TH2 cells producing some or all of IL-4, IL-5,

S. Crotty





IL-13, and IL-10), the vastness of the dimen-sional space of memory T-cell phenotypic het-erogeneity was first made evident by the humanmemory CD8 T-cell mass spectrometry study ofNewell and Davis (Newell et al. 2012). There wassuch phenotypic variety of memory CD8 T cellsto influenza and cytomegalovirus (CMV) thatthe authors did not even attempt to put a num-ber on the total range of memory CD8 T-cellphenotypes observed; instead, the diversity wasbest calculated as large spaces of phenotypicvariation in three-dimensional principal com-ponent analysis (PCA) plots (Newell et al.2012). Subsequent mass spectrometry analysisof human CD4 T cells has shown even moreheterogeneity (Wong et al. 2015), consistentwith the diversity of TH1, TH2, TH17, TFH,TH9, TH1�, CD4 CTL, and iTreg biology. Impor-tantly, phenotypic diversity in human memoryCD4 T cells is seen even at the level of individualTCR clonotypes (Becattini et al. 2015). One ex-ample of heterogeneity is the presence or ab-sence of PD-1 expression by resting memoryTFH cells (Locci et al. 2013). Heterogeneity isclearly present among chemokine receptor ex-pression by memory CD4 T cells. A populationof CCR6þCXCR5þ resting memory CD4 T cellsis present in human peripheral blood, and it hasbeen suggested those cells represent “TFH17”memory cells (Morita et al. 2011). However,CCR6 expression is not specific to TH17 cellsand does not correlate well with TH17 program-ming in many cases. One report found no IL-17a expression by stimulated CCR6þ CXCR5þ

memory CD4 cells in single-cell analysis (Wonget al. 2015). Therefore, most CCR6 expressionby memory TFH cells may be unrelated to TH17biology and simply reflective of preferentialchemotaxis needs. A similar situation existsfor “TFH2” memory cells. “TFH2” memoryCD4 T cells in human peripheral blood arecommonly defined only on the basis of negativemarkers (CXCR32CCR62) (Morita et al. 2011;Jacquemin et al. 2015), based on the assump-tion that all TFH memory cells must have a TH1,TH2, or TH17 bias (which has certainly notbeen demonstrated at the single-cell level forGC TFH cells or resting memory TFH cells);thus, the concept of TFH2 cells remains poorly

defined. Given that TFH cells produce IL-4 in aTFH-specific manner (independent of the TH2program) (Crotty 2011), identification of aTFH2 cell requires demonstration of a restingmemory CXCR5þ CD4 T cell capable ofproducing IL-5 and/or IL-13 at the single-celllevel—a criterion not reached in the currentliterature. Such cells may exist, but they havenot been shown directly, and they may be aminor fraction of the CXCR32CCR62 memoryTFH cells. There is clearly vast phenotypic het-erogeneity among resting memory CD4 T cells,including resting memory TFH cells. However,at the level of cytokine production, it is stillunknown how rare memory TFH cells with cy-tokine production attributes of other CD4 T-cell subsets are, except for overlap between TH1and TFH programs in memory CD4 T cells,which has been clearly shown by multiplegroups.

PLASTICITY WHEN MEMORY CD4 T CELLSARE RECALLED?

A first study using an IL-21-GFP reportermouse observed extensive plasticity of IL-21-GFPþ CXCR5þ TFH or IL-21-GFP2 CXCR5þ

cells after transfer into new hosts, with ,50%of the memory cells maintaining CXCR5 ex-pression, and a majority of the cells observedafter recall by influenza infection wereCXCR52 (Luthje et al. 2012). A newer IL-21fluorescent protein reporter mouse model ob-served robust stability of IL-21+ CXCR5+ TFH

CD4 T cells after transfers into new hosts, butdid not test memory time points (Weinsteinet al. 2016). In the context of an acute LCMVinfection, memory CD4 T cells appear to largelymaintain their TH1 or TFH programming upon2˚ response. TCR transgenic memory CD4 Tcells with a resting TH1 phenotype all becameeffector TH1 cells (T-bethiBcl6loCXCR52

Granzyme Bþ) when transferred into a newhost that was then infected with LCMV. In thesame study, the major of memory CD4 T cellswith a resting TFH phenotype became effectorTFH and GC TFH cells on rechallenge (T-betloBcl6þCXCR5þGranzyme B2). A fractionof the memory TFH cells did lose CXCR5 in






the 2˚ response, but because those experimentsdepended on cell transfers, it is unknownwhether all of the memory TFH cells wouldhave maintained their TFH program on 2˚ chal-lenge under physiological conditions in an in-tact animal where they were allowed to maintaintheir normal localization. It is, of course, alsoformally possible that the memory cells wouldhave showed even more plasticity if studying inan intact host. Another longstanding challengewith sorted cells that have high proliferative ca-pacity such as lymphocytes is that even a 1%CXCR52 contamination of sorted cells couldsubsequently expand extensively during the ex-ponential proliferation following 2˚ challenge,thus confounding cell-fate interpretations ofcell-transfer experiments.

Resting central memory CD4 T cells showless polarized features than actively respondingeffector cells. Resting central memory TH1 orTFH cells have less active transcription andprotein expression of many signature featuresof activated TH1 cells or activated GC TFH cells.For example, central memory TH1 cells expresslow levels of T-bet compared with effector cells,and resting memory TFH cells express Bcl6 butat levels indistinguishable from other centralmemory cells. Intuitively, one expects suchcells to be prone to plasticity. Programmedchromatin modifications may prevent plasticity.Memory TFH cells rapidly up-regulate Bcl6 invivo on restimulation (Ise et al. 2014). It is un-known whether memory TFH cells with mixedTFH–TH1 phenotypes (CXCR3þCXCR5þ)differentiate into TFH (CXCR32CXCR5þBcl6þ

T-bet2), TFH1 (CXCR3þCXCR5þBcl6þT-betþ),and/or conventional TH1 cells (CXCR3þ

CXCR52Bcl62T-betþ) in vivo in an intactmouse or human (Fig. 1). Evidence of plasticitywas not observed in humans at the level of theoverall CD4 T-cell response to pertussis, where-in the whole-cell pertussis vaccine and the acel-lular pertussis vaccine elicit predominantly TH1and TH2 polarizing responses, respectively, andreimmunization with the acellular pertussisvaccine elicits a CD4 T-cell recall responsewith the same TH1 or TH2 characteristics forwhichever vaccine the individual was initiallyimmunized (Bancroft et al. 2016).

CONCLUDING REMARKS

There remains no direct demonstration of invivo–generated resting memory CD4 T-cellconversion to a different subtype on secondaryantigen challenge in vivo in an intact animal atthe single-cell level. Lineage tracing experimentsare needed to directly test whether plasticityoccurs and, more importantly, how commonor rare the process is under physiological con-ditions. Unfortunately, proper lineage-trackinggenetic models are difficult to generate for TFH,TH17, and TH2 cells. The lineage-definingtranscription factors for TFH, TH17, and TH2cells are Bcl6, RORgT, and GATA3. GATA3 isconstitutively expressed by all CD4 T cells; TH2cells are defined by high GATA3 expression.Thus, a conventional GATA3 lineage reporterwould mark all T cells. Therefore, developmentof a high-fidelity TH2 lineage genetic marker isa difficult challenge. A successful TH2 lineagegenetic marker would probably need to de-pend on transcription from a gene or locusother than GATA3. Bcl6 is highly expressedby thymocytes, and thus a Bcl6 lineage report-er mouse constructed based on standard de-signs would be expected to mark most T cells.RORgt is also expressed by thymocytes and,thus, an RORgt lineage reporter mouse basedon standard designs would be expected tomark most T cells. Therefore, a successfulTFH or TH17 lineage genetic marker wouldprobably need to depend on transcriptionfrom a gene or locus other than GATA3; IL-17a is one candidate for TH17 cells (Hirotaet al. 2011). CXCR5 may be a good candidatefor TFH cells. As for TH1 cells, a T-bet lineagereporter should be useful, but a key caveat isthat transient expression of T-bet early on by anaıve CD4 T cell after activation is commonand may have no influence on the future his-tory of the cell. The same caveat applies for allconstitutively active lineage marker systems,and contributed to controversy over the ontog-eny of “exTregs.” Transient expression of Foxp3by some cells may result in erroneous conclu-sions on the basis of very transient Cre expres-sion (Rubtsov et al. 2008; Zhou et al. 2009;Miyao et al. 2012). This concern may be

S. Crotty





avoided by using estrogen-responsive Cre pro-tein (Rubtsov et al. 2010).

As an alternative to lineage-tracking geneticmarkers, single-cell transfers into infection-matched hosts may be the best test of stabilityversus plasticity in a recall response. Such ex-periments would need to show that transferredcells (not using single-cell transfers) would lo-calize in the new host to the same regions oflymph node (LN) and spleen as untransferredantigen-specific resting memory CD4 T cells ofthat subtype (e.g., proper microanatomical lo-calization of memory TFH cells posttransfer).

In the end, it is unknown whether the ap-pearance of plasticity by memory CD4 T cells atthe population level is predominantly because ofheterogeneity and outgrowth of subpopulationsor predominantly attributable to plasticity.

It remains possible that true memory CD4T-cell plasticity may be primarily of interest forpurely academic reasons, as a memory recallresponse with apparent plasticity at the whole-cell population level could presumably be ac-complished via rapid outgrown of a very minorpopulation of memory cells. Because essentiallyall in vivo CD4 T-cell responses involve a mix-ture of subtypes (TH1, TH2, TH CTL, and TFH,for example), that scenario is plausible.

ACKNOWLEDGMENTS

This work is funded by the National Institutes ofHealth (NIH) National Institute of Allergy andInfectious Diseases (NIAID) R01 (S.C.). Thanksto Daniel DiToro for helpful discussions.

REFERENCES

Ballesteros-Tato A, Randall TD, Lund FE, Spolski R, Leo-nard WJ, Leon B. 2016. T follicular helper cell plasticityshapes pathogenic T helper 2 cell-mediated immunity toinhaled house dust mite. Immunity 44: 259–273.

Bancroft T, Dillon MBC, Da Silva Antunes R, Paul S, PetersB, Crotty S, Lindestam Arlehamn CS, Sette A. 2016. Th1versus Th2 T cell polarization by whole-cell and acellularchildhood pertussis vaccines persists upon re-immuni-zation in adolescence and adulthood. Cell Immunol 304 –305: 35–43.

Becattini S, Latorre D, Mele F, Foglierini M, De Gregorio C,Cassotta A, Fernandez B, Kelderman S, Schumacher TN,Corti D, et al. 2015. Functional heterogeneity of human

memory CD4þ T cell clones primed by pathogens orvaccines. Science 347: 400–406.

Bentebibel SE, Lopez S, Obermoser G, Schmitt N, MuellerC, Harrod C, Flano E, Mejias A, Albrecht RA, Blanken-ship D, et al. 2013. Induction of ICOSþCXCR3þCXCR5þ

TH cells correlates with antibody responses to influenzavaccination. Sci Transl Med 5: 176ra32.

Choi YS, Yang JA, Yusuf I, Johnston RJ, Greenbaum J, PetersB, Crotty S. 2013. Bcl6 expressing follicular helper CD4T cells are fate committed early and have the capacity toform memory. J Immunol 190: 4014–4026.

Coquet JM, Schuijs MJ, Smyth MJ, Deswarte K, Beyaert R,Braun H, Boon L, Karlsson Hedestam GB, Nutt SL, Ham-mad H, et al. 2015. Interleukin-21-producing CD4þ

T cells promote type 2 immunity to house dust mites.Immunity 43: 318–330.

Crotty S. 2011. Follicular helper CD4 T cells (TFH). AnnuRev Immunol 29: 621–663.

Crotty S. 2012. The 1-1-1 fallacy. Immunol Rev 247: 133–142.

Crotty S. 2014. T follicular helper cell differentiation, func-tion, and roles in disease. Immunity 41: 529–542.

Dan JM, Lindestam Arlehamn CS, Weiskopf D, dda SilvaAntunes R, Havenar-Daughton C, Reiss SM, Brigger M,Bothwell M, Sette A, Crotty, S. et al.2016. A cytokine-independent approach to identify antigen-specific hu-man germinal center T follicular helper cells and rareantigen-specific CD4þ T cells in blood. J Immunol 197:983–993.

Da Silva Antunes R, Paul S, Sidney J, Weiskopf D, Dan JM,Phillips E, Mallal S, Crotty S, Sette A, Arlehamn CSL.2017. Definition of human epitopes recognized in teta-nus toxoid and development of an assay strategy to detectex vivo tetanus CD4þ T cell responses. PLoS ONE 12:e0169086.

DuPage M, Bluestone JA. 2016. Harnessing the plasticity ofCD4þ T cells to treat immune-mediated disease. Nat RevImmunol 16: 149–163.

Hale JS, Youngblood B, Latner DR, Mohammed AUR, Ye L,Akondy RS, Wu T, Iyer SS, Ahmed R. 2013. Distinctmemory CD4þ T cells with commitment to T follicularhelper- and T helper 1-cell lineages are generated afteracute viral infection. Immunity 38: 805–817.

Harrington LE, Janowski KM, Oliver JR, Zajac AJ, WeaverCT. 2008. Memory CD4 T cells emerge from effectorT-cell progenitors. Nature 452: 356–360.

Hatzi K, Nance JP, Kroenke MA, Bothwell M, Haddad EK,Melnick A, Crotty S. 2015. BCL6 orchestrates Tfh celldifferentiation via multiple distinct mechanisms. J ExpMed 212: 539–553.

Havenar-Daughton C, Reiss SM, Carnathan DG, Wu JE,Kendric K, Torrents de la Pena A, Kasturi SP, Dan JM,Bothwell M, Sanders RW, et al. 2016. Cytokine-indepen-dent detection of antigen-specific germinal center T fol-licular helper cells in immunized nonhuman primatesusing a live cell activation-induced marker technique.J Immunol 197: 994–1002.

He J, Tsai LM, Leong YA, Hu X, Ma CS, Chevalier N, Sun X,Vandenberg K, Rockman S, Ding Y, et al. 2013. Circulat-ing precursor CCR7loPD-1hi CXCR5þ CD4þ T cells in-dicate Tfh cell activity and promote antibody responsesupon antigen reexposure. Immunity 39: 770–781.






Hirota K, Duarte JH, Veldhoen M, Hornsby E, Li Y, Cua DJ,Ahlfors H, Wilhelm C, Tolaini M, Menzel U, et al. 2011.Fate mapping of IL-17-producing T cells in inflammato-ry responses. Nat Immunol 12: 255–263.

Hondowicz BD, An D, Schenkel JM, Kim KS, Steach HR,Krishnamurty AT, Keitany GJ, Garza EN, Fraser KA,Moon JJ, et al. 2016. Interleukin-2-dependent allergen-specific tissue-resident memory cells drive asthma. Im-munity 44: 155–166.

Ise W, Inoue T, McLachlan JB, Kometani K, Kubo M, OkadaT, Kurosaki T. 2014. Memory B cells contribute to rapidBcl6 expression by memory follicular helper T cells. ProcNatl Acad Sci 111: 11792–11797.

Iyer SS, Gangadhara S, Victor B, Gomez R, Basu R, Hong JJ,Labranche C, Montefiori DC, Villinger F, Moss B, et al.2015. Codelivery of envelope protein in alum with MVAvaccine induces CXCR3-biased CXCR5þ and CXCR52

CD4 T cell responses in rhesus macaques. J Immunol 195:994–1005.

Jacquemin C, Schmitt N, Contin-Bordes C, Liu Y, Nara-yanan P, Seneschal J, Maurouard T, Dougall D, DavizonES, Dumortier H, et al. 2015. OX40 ligand contributes tohuman lupus pathogenesis by promoting T follicularhelper response. Immunity 42: 1159–1170.

Johnston RJ, Poholek AC, DiToro D, Yusuf I, Eto D, BarnettB, Dent AL, Craft J, Crotty S. 2009. Bcl6 and Blimp-1 arereciprocal and antagonistic regulators of T follicularhelper cell differentiation. Science 325: 1006–1010.

Kitano M, Moriyama S, Ando Y, Hikida M, Mori Y, KurosakiT, Okada T. 2011. Bcl6 protein expression shapes pre-germinal center B cell dynamics and follicular helperT cell heterogeneity. Immunity 34: 961–972.

Liang HE, Reinhardt RL, Bando JK, Sullivan BM, Ho IC,Locksley RM. 2012. Divergent expression patterns of IL-4and IL-13 define unique functions in allergic immunity.Nat Immunol 13: 58–66.

Lindenstrøm T, Woodworth J, Dietrich J, Aagaard C, Ander-sen P, Agger EM. 2012. Vaccine-induced th17 cells aremaintained long-term postvaccination as a distinct andphenotypically stable memory subset. Infect Immun 80:3533–3544.

Liu X, Yan X, Zhong B, Nurieva RI, Wang A, Wang X, Mar-tin-Orozco N, Wang Y, Chang SH, Esplugues E, et al.2012. Bcl6 expression specifies the T follicular helpercell program in vivo. J Exp Med 209: 1841–1852.

Locci M, Havenar-Daughton C, Landais E, Wu J, KroenkeMA, Arlehamn CL, Su LF, Cubas R, Davis MM, Sette A,et al. 2013. Human circulating PD-1þCXCR3-CXCR5þ

memory Tfh cells are highly functional and correlatewith broadly neutralizing HIV antibody responses. Im-munity 39: 758–769.

Lu KT, Kanno Y, Cannons JL, Handon R, Bible P, ElkahlounAG, Anderson SM, Wei L, Sun H, O’Shea JJ, et al. 2011.Functional and epigenetic studies reveal multistep differ-entiation and plasticity of in vitro–generated and invivo–derived follicular T helper cells. Immunity 35:622–632.

Luthje K, Kallies A, Shimohakamada Y, Belz GT, Light A,Tarlinton DM, Nutt SL. 2012. The development and fateof follicular helper T cells defined by an IL-21 reportermouse. Nat Immunol 13: 491–498.

Ma CS, Suryani S, Avery DT, Chan A, Nanan R, Santner-Nanan B, Deenick EK, Tangye SG. 2009. Early commit-ment of naıve human CD4þ T cells to the T follicularhelper (TFH) cell lineage is induced by IL-12. ImmunolCell Biol 87: 590–600.

McGeachy MJ. 2013. Th17 memory cells: Live long andproliferate. J Leukoc Biol 94: 921–926.

Miyao T, Floess S, Setoguchi R, Luche H, Fehling HJ, Wald-mann H, Huehn J, Hori S. 2012. Plasticity of Foxp3þ

T cells reflects promiscuous Foxp3 expression in conven-tional T cells but not reprogramming of regulatory Tcells. Immunity 36: 262–275.

Morita R, Schmitt N, Bentebibel SE, Ranganathan R, Bour-dery L, Zurawski G, Foucat E, Dullaers M, Oh S, Sabz-ghabaei N, et al. 2011. Human blood CXCR5þCD4þ Tcells are counterparts of T follicular cells and containspecific subsets that differentially support antibody secre-tion. Immunity 34: 108–121.

Muranski P, Borman ZA, Kerkar SP, Klebanoff CA, Ji Y,Sanchez-Perez L, Sukumar M, Reger RN, Yu Z, Kern SJ,et al. 2011. Th17 cells are long lived and retain a stem cell-like molecular signature. Immunity 35: 972–985.

Nakayamada S, Kanno Y, Takahashi H, Jankovic D, Lu KT,Johnson TA, Sun HW, Vahedi G, Hakim O, Handon R, etal. 2011. Early Th1 cell differentiation is marked by a Tfhcell-like transition. Immunity 35: 919–931.

Newell EW, Sigal N, Bendall SC, Nolan GP, Davis MM. 2012.Cytometry by time-of-flight shows combinatorial cyto-kine expression and virus-specific cell niches within acontinuum of CD8þ T cell phenotypes. Immunity 36:142–152.

Obeng-Adjei N, Portugal S, Tran TM, Yazew TB, Skinner J,Li S, Jain A, Felgner PL, Doumbo OK, Kayentao K, et al.2015. Circulating Th1-cell-type Tfh cells that exhibit im-paired B cell help are preferentially activated during acutemalaria in children. Cell Rep 13: 425–439.

Oestreich KJ, Mohn SE, Weinmann AS. 2012. Molecularmechanisms that control the expression and activity ofBcl-6 in TH1 cells to regulate flexibility with a TFH-likegene profile. Nat Immunol 13: 405–411.

Pepper M, Linehan JL, Pagan AJ, Zell T, Dileepan T, ClearyPP, Jenkins MK. 2010. Different routes of bacterial infec-tion induce long-lived TH1 memory cells and short-livedTH17 cells. Nat Immunol 11: 83–89.

Ray JP, Staron MM, Shyer JA, Ho P-C, Marshall HD, GraySM, Laidlaw BJ, Araki K, Ahmed R, Kaech SM, et al. 2015.The interleukin-2-mTORc1 kinase axis defines the sig-naling, differentiation, and metabolism of T helper 1 andfollicular B helper T cells. Immunity 43: 690–702.

Rubtsov YP, Rasmussen JP, Chi EY, Fontenot J, Castelli L, YeX, Treuting P, Siewe L, Roers A, Henderson WR, et al.2008. Regulatory T cell-derived interleukin-10 limits in-flammation at environmental interfaces. Immunity 28:546–558.

Rubtsov YP, Niec RE, Josefowicz S, Li L, Darce J, Mathis D,Benoist C, Rudensky AY. 2010. Stability of the regulatoryT cell lineage in vivo. Science 329: 1667–1671.

Tubo NJ, Pagan AJ, Taylor JJ, Nelson RW, Linehan JL, ErteltJM, Huseby ES, Way SS, Jenkins MK. 2013. Single naiveCD4þ T cells from a diverse repertoire produce differenteffector cell types during infection. Cell 153: 785–796.

S. Crotty





Tubo NJ, Fife BT, Pagan AJ, Kotov DI, Goldberg MF, JenkinsMK. 2016. Most microbe-specific naıve CD4þ T cells pro-duce memory cells during infection. Science 351: 511–514.

Vinuesa CG, Linterman MA, Yu D, Maclennan ICM. 2016.Follicular helper T cells. Annu Rev Immunol 34: 335–368.

Weinstein JS, Herman EI, Lainez B, Licona-Limon P, Es-plugues E, Flavell R, Craft J. 2016. TFH cells progressivelydifferentiate to regulate the germinal center response. NatImmunol 17: 1197–1205.

Wong MT, Chen J, Narayanan S, Lin W, Anicete R, KiaangHTK, De Lafaille MAC, Poidinger M, Newell EW. 2015.Mapping the diversity of follicular helper T cells in hu-man blood and tonsils using high-dimensional mass cy-tometry analysis. Cell Rep 11: 1822–1833.

Yu D, Rao S, Tsai LM, Lee SK, He Y, Sutcliffe EL, SrivastavaM, Linterman M, Zheng L, Simpson N, et al. 2009. The

transcriptional repressor Bcl-6 directs T follicular helpercell lineage commitment. Immunity 31: 457–468.

Yusuf I, Kageyama R, Monticelli L, Johnston RJ, DiToro D,Hansen K, Barnett B, Crotty S. 2010. Germinal centerT follicular helper cell IL-4 production is dependenton signaling lymphocytic activation molecule receptor(CD150). J Immunol 185: 190–202.

Zhou X, Bailey-Bucktrout SL, Jeker LT, Penaranda C, Mar-tınez-Llordella M, Ashby M, Nakayama M, Rosenthal W,Bluestone JA. 2009. Instability of the transcription factorFoxp3 leads to the generation of pathogenic memoryT cells in vivo. Nat Immunol 10: 1000–1007.

Zielinski CE, Mele F, Aschenbrenner D, Jarrossay D, RonchiF, Gattorno M, Monticelli S, Lanzavecchia A, Sallusto F.2012. Pathogen-induced human TH17 cells produceIFN-g or IL-10 and are regulated by IL-1b. Nature 484:514–518.






do memory cd4 t cells keep their cell-type programming ... · cellular reprogramming techniques,...

Documents